101. Development of an analytical pipeline for detecting fusions of clinical relevance in liver cancer

Abstract

Gene fusions play a crucial role in multiple cancers including cholangiocarcinoma (CCA) and hepatocellular carcinoma (HCC). To make unbiased gene-fusion discoveries for CCA and HCC, we studied tumor RNA-seq data from 51 CCA and 48 HCC patients treated at Mayo Clinic, and an additional 424 HCC and CCA cases from The Cancer Genome Atlas (TCGA). We developed an integrative fusion pipeline to annotate and identify high confidence, candidate gene fusions initially called by STAR-Fusion-v1.4.0. The pipeline considers exon boundaries, in-frame candidates, homologous genes, and the proximity of gene partners. The pipeline removes fusions with low read support, artifacts found in normal tissues, or fusions present within genes known to be irrelevant in cancer. Among the highly confident fusion calls (>50 supporting reads) recurrent across multiple tumors, we identified DNAJB1-PRKACA, characteristic of fibrolamellar hepatocellular carcinoma (FL-HCC) in two unrelated HCC samples. This fusion was also found in six TCGA samples previously determined to have FL-HCC through an independent pathological evaluation, blind to the RNA-seq results. FL-HCC is known to primarily affect young adults, all eight cases with detected DNAJB1-PRKACA fusions were diagnosed at a young age (17-34 years old). In six unrelated CCA samples, we identified multiple FGFR2 gene fusions paired with various fusion partners, including CCAR1 a tumor-related regulator. These bioinformatics findings support the technical readiness and discovery potential of our fusion pipeline in studying HCC and CCA. Other novel and potentially important oncogenic fusions were identified, including SMAD. Future validation will be performed to confirm all key findings. Gene fusions play a crucial role in multiple cancers including cholangiocarcinoma (CCA) and hepatocellular carcinoma (HCC). To make unbiased gene-fusion discoveries for CCA and HCC, we studied tumor RNA-seq data from 51 CCA and 48 HCC patients treated at Mayo Clinic, and an additional 424 HCC and CCA cases from The Cancer Genome Atlas (TCGA). We developed an integrative fusion pipeline to annotate and identify high confidence, candidate gene fusions initially called by STAR-Fusion-v1.4.0. The pipeline considers exon boundaries, in-frame candidates, homologous genes, and the proximity of gene partners. The pipeline removes fusions with low read support, artifacts found in normal tissues, or fusions present within genes known to be irrelevant in cancer. Among the highly confident fusion calls (>50 supporting reads) recurrent across multiple tumors, we identified DNAJB1-PRKACA, characteristic of fibrolamellar hepatocellular carcinoma (FL-HCC) in two unrelated HCC samples. This fusion was also found in six TCGA samples previously determined to have FL-HCC through an independent pathological evaluation, blind to the RNA-seq results. FL-HCC is known to primarily affect young adults, all eight cases with detected DNAJB1-PRKACA fusions were diagnosed at a young age (17-34 years old). In six unrelated CCA samples, we identified multiple FGFR2 gene fusions paired with various fusion partners, including CCAR1 a tumor-related regulator. These bioinformatics findings support the technical readiness and discovery potential of our fusion pipeline in studying HCC and CCA. Other novel and potentially important oncogenic fusions were identified, including SMAD. Future validation will be performed to confirm all key findings.